JP6960502B2 - Copper pillars with improved integrity and methods of manufacturing them - Google Patents

Description

The present invention is directed to copper pillars with improved integrity and methods of making copper pillars. More specifically, the present invention relates to copper pillars with improved integrity and methods of manufacturing copper pillars such that the copper pillars do not easily break after solder bump application and reflow.

Copper pillars are photoresist defined features for integrated circuit chips and printed circuit boards. The features are formed by a lithography process in which the photoresist is applied to a semiconductor wafer chip, often referred to as a die in packaging technology, or a substrate such as an epoxy / glass printed circuit board. The photoresist is applied to the surface of the substrate, and a mask with a pattern is applied to the photoresist. The substrate with the mask is exposed to radiation such as UV light. The radiation-exposed photoresist compartments are developed or removed to expose the surface of the substrate. The contours of the openings are formed with the unexposed photoresist remaining on the substrate forming the walls of the openings. The surface of the substrate comprises a metal seed layer or other conductive metal or metal alloy material that can make the surface of the substrate conductive. The substrate with the patterned photoresist is then immersed in a copper electroplating bath, and copper is electroplated into the openings to form pillars. When electroplating is complete, the rest of the photoresist is stripped from the substrate with a stripping solution to further process the substrate with the photoresist-defined features.

Copper pillars are typically solder capped to allow adhesion and electrical conduction between the semiconductor chip on which the pillars are plated and the substrate. Such sequences are found in advanced packaging techniques. Solder-capped copper pillar structures are a rapidly growing part in advanced packaging applications due to the improved input / output (I / O) densities compared to solder bumping alone. Copper pillars with a structure of non-reflowable copper pillars and reflowable solder bumps have the following advantages: (1) copper has low electrical resistance and high current density capability, (2) copper thermal conductivity. Provides more than three times the thermal conductivity of solder bumps, (3) can cause reliability problems, can remedy the problem of conventional BGA CTE (coefficient of thermal expansion) mismatch, and ( 4) The copper pillars do not collapse during reflow, enabling very fine pitches without compromising the standoff height.

Of all the copper pillar bump manufacturing processes, electroplating is by far the most commercially viable process. In actual industrial production, considering costs and process conditions, electroplating provides mass productivity and there is no polishing or corrosion process that changes the surface morphology of the copper pillars after they are formed. The ideal copper electroplating chemistry and method for electroplating copper pillars is a deposit with good uniformity, preferably a flat or dish-shaped top and a void-free metal-to-metal interface after reflowing with solder. Bring. Flat or dish-shaped pillar tops are generally preferred in the industry, as solder bumps applied to the dome-shaped top pillars tend to run off the top along the sides of the pillars during reflow. In addition, it is highly preferred in the industry to plate with a high deposition rate to allow for high wafer through-out. High deposition rates typically exceed 10 ASD, more typically 15 ASD, and even more typically 20 ASD and above.

While forming copper pillars by electroplating and electroplating tin or tin alloy solder bumps on copper pillars, copper pillars and solder bumps, in particular copper-tin, where the copper pillars join to tin or tin alloy solder bumps. At the metal-to-metal interface, it is exposed to a number of different stresses that can lead to damage to the copper pillars. Voids, often referred to as Kirkendal voids, are formed within the copper-tin metal interface. These voids can have very small diameters, but when many between them are formed within the metal-to-metal interface, they coalesce into larger voids and crack, forming copper pillars and solder bumps. It results in solder bump electrical resistance that impairs electrical performance.

An example of a conventional process for plating copper pillars is shown in FIGS. 1A-1D, wherein a semiconductor substrate 10 such as a silicon wafer having a seed layer (not shown) such as a copper seed layer is provided. Will be done. The semiconductor substrate is coated with a photoresist layer 12 such as a positive photoresist, and a mask having an aperture pattern (not shown) is applied on top of the photoresist layer. Irradiate the photoresist (not shown) with UV light. The portion of the photoresist exposed to UV light through the pattern of openings becomes soluble and removed in the developer, thereby forming a series of openings 14 and 16 through the photoresist exposing the seed layer on the semiconductor substrate. The substrate with the patterned photoresist is then immersed in a copper electroplating bath where a plurality of domed copper pillars 18 and 20 are plated in openings 14 and 16. The current density during copper electroplating typically exceeds 10 ASD, and more typically the current density is about 20 ASD. Electroplating is conformal or same plating rate precipitation everywhere, not super conformal or super filling. The tin or tin alloy solder bumps 22 and 24 are then deposited on the top of each dome-shaped pillar by electroplating, physical or chemical deposition, or the like. Solder bumps are usually electroplated on the tops of copper pillars 18 and 20. As shown in FIG. 1C, due to the copper pillars having a domed or convex shape, the solder bumps are unnecessarily plated along the sides of the pillars, not only at the top of the pillars but also irregularly 15 and 17. Tend to do. Following the deposition of solder bumps, the photoresist 12 is removed from the semiconductor substrate with a stripping solvent, leaving copper pillars 18 and 20 with solder bumps 22 and 24 on the semiconductor substrate 10. Next, the semiconductor substrate having the copper pillars and the solder bumps is reflowed in the reflow oven. After reflow, a copper-tin interface 25 with a significant number of voids 26 is formed between the solder bumps and the copper pillars. As mentioned above, such voids can cause cracks in the copper pillar and solder bump structures at the interface 25, thus resulting in defective and damaged electrical equipment.

The development of plating chemicals and methods to reduce the number of voids at the interface between solder bumps and copper pillars and reduce the potential for cracks is very difficult for the industry. Structures based on copper pillars have already been used by various manufacturers for use in consumer products such as smartphones and PCs. As wafer level processing (WLP) evolves and continues to adopt the use of copper pillar technology, the demand for reliable copper pillar construction will increase.

Therefore, there is a need for copper pillars with improved structural integrity and methods for producing them.

The present invention is a copper pillar comprising a horizontal base and a vertical section including a top side opposite to the bottom side and a spare bottom side, wherein the bottom side of the vertical section is joined to the horizontal base. A tin or tin alloy solder bump is joined to the top side of the vertical section by a copper-tin metal interface, said 5 μm from the copper of the copper pillar along the length of the copper-tin metal interface. The target is copper pillars with V% = 20% or less, up to the inside between copper and tin metal.

The present invention also
a) A substrate including a photoresist layer, wherein the photoresist layer includes a plurality of openings.
b) A first containing one or more copper ion sources, one or more acids, one or more chloride sources, one or more levelers, one or more accelerators, and one or more inhibitors. Preparing a copper electroplating bath and
c) Immersing a substrate containing a photoresist layer having a plurality of openings in a first copper electroplating bath, and
d) Electroplating the first section of copper pillars into each of the openings at a first current density, followed by a plurality of copper electroplating baths at a current density lower than the first current density. Each of the openings is electroplated with a second section of copper pillars, or water, one or more copper ion sources, one or more acids, and optionally one or more chloride sources, one or more. A second copper electroplating bath consisting of a leveler, one or more accelerators, and one or more inhibitors, electroplating a second section of copper pillars into each of a plurality of openings at a lower current density. That and
e) Depositing tin or tin alloy solder bumps on the top of the second section of each copper pillar
f) Peeling the photoresist from the substrate, leaving an array of copper pillars with tin or tin alloy solder bumps on the top of the second section of each copper pillar.
g) Includes reflowing the array and electroplating copper pillars, including.

The copper pillars of the present invention have improved integrity over many conventional copper pillars. The number of voids in the copper-tin metal interface that joins the copper of the copper pillars to the solder bumps of the tin or tin alloy is substantially reduced after reflow, and the copper-tin metal interface is significantly damaged or cracked. Reduced or prevented. The reduction in the number of voids in the copper-tin metal interface reduces the likelihood that adjacent voids will coalesce into larger voids, thereby preventing cracks at the interface and separation of solder bumps from the copper pillars. ..

An explanatory view of a conventional process for plating a dome-shaped copper pillar, the copper pillar comprises a solder bump and a copper-tin metal interface with significant void formation. FIG. 5 is a view of a copper pillar of the invention having a flat top with solder bumps, a dish-shaped top and a dome-shaped top and a substantially void-free copper-tin metal interface on the top section of each copper pillar. It is a figure of the copper-tin metal interface adjacent to the top of the flat top copper pillar of the present invention, and the metal interface along the length is substantially void-free. FIG. 5 is a diagram of a copper-tin metal interface adjacent to the top of a conventional copper pillar with a flat top, the metal interface along that length containing a significant number of voids of various lengths (diameters). FIG. 5 is a diagram of a method of making an array of copper pillars of the present invention, where the copper pillars have a flat top and a substantially voidless copper-tin metal interface along their length. FIG. 5 is an array of pre-reflow and post-reflow countersunk copper pillars with solder bumps, the reflow copper pillars containing a substantially voidless copper-tin metal interface along their length. The SEM is 3,500 times the cross section of a copper pillar having a dish-shaped top and a copper-tin metal interface of the present invention, and the metal interface contains very few voids with a very small diameter or is a metal interface. Is virtually void-free. An SEM of 3,500 times the cross section of a comparative copper pillar having a dish-shaped top and a copper-tin metal interface with substantial voids in and along the length of the interface. ..

As used throughout this specification, the following abbreviations shall have the following meanings, unless the context clearly indicates otherwise: A = amps, A / dm ² = amps / square. Decimetre = ASD, ° C = degrees Celsius, UV = ultraviolet rays, g = grams, ppm = millions = mg / L, L = liters, μm = microns = micrometers, mm = millimeters, cm = centimeters, DI = Deionized, mL = milliliter, mol = mol, mmol = mmol, sec = seconds, C = cycle, Mw = weight average molecular weight, Mn = number average molecular weight, SEM = scanning electron microscope, FIB = focused ion beam , TIR = total indicator runout = maximum indicator reading = full indicator movement = FIM, and Avg. = Average, V% =% void formation, EO = ethylene oxide, PO = propylene oxide, IMC = intermetallic compound, Cu-Cu ₃ Sn IMC = copper-tin intermetallic compound, and NIH = National Institute of Health (National Institutes of Health) Place).

As used throughout this specification, the term "plating" refers to metal electroplating. "Deposition" and "plating" are used interchangeably throughout this specification. "Accelerator" refers to an organic additive that increases the plating rate of an electroplating bath. "Inhibitor" refers to an organic additive that suppresses the plating rate of metals during electroplating. The term "array" means a regular array. The term "opening" means an opening or hole. Term "dome shape" = convex. Term "dish-shaped" = concave. The term "form" means the form, shape, and structure of an article. The term "runout accuracy" or "maximum reading difference" is the difference between the maximum and minimum readings of a measured value, that is, an indicator on a plane, cylinder, or contoured surface of a part. , Indicates the amount of deviation from flatness, roundness (roundness), cylindricity, and concentricity with other cylindrical features or similar conditions. The term "profilimetry" means the use of techniques in the measurement and profiling of objects, or the use of lasers or white light computer-generated projections to make surface measurements of three-dimensional objects. The term "normalized" means rescaling to reach values associated with size variables such as ratio as TIR%. The term "mean" is the mean of the parameters, where "mean" is the sum of all samples of the measured or determined parameter divided by the total number of samples. The term "parameter" means a numerical factor or other measurable factor that forms one of a set that defines a system or sets conditions for its operation. The articles "a" and "an" refer to the singular and plural.

All numerical ranges are inclusive and can be combined in any order, except where it is clear that such numerical ranges are limited to a maximum of 100%.

The present invention covers the copper pillar 100, wherein the copper pillar includes a horizontal base 102, a vertical section 104, and a top side 106 opposite the horizontal base. In the vertical section 104, both the top side 106 and the horizontal base 102 coincide with each other to form a single copper pillar structure. The top side can be flat and parallel to the horizontal base as shown in FIG. 2A, or the top side can be dish-shaped as shown in FIG. 2B, or the top side can be dome-shaped as shown in FIG. 2C. obtain.

Preferably, the top 106 of the copper pillars is flat or dish-shaped, more preferably the top 106 is dish-shaped as shown in FIG. 2B. When the top side is flat, the TIR% is preferably in the range of -1% to + 2%, more preferably 0% to + 2%, and even more preferably 0% to + 1%. When the top side is dish-shaped, the TIR% is preferably in the range of -10% to less than -1%, more preferably -10% to -7%, and even more preferably -9% to -7%. .. When the top of the copper pillar is domed, the TIR% is preferably in the range of more than + 2% to + 10%, more preferably + 2.5% to + 10%, even more preferably + 5% to + 10%.

The TIR% of the copper pillar can be determined by the following formula.
TIR% = [Height _center -Height _side ] / _Maximum height x 100
Wherein the height _center is the height of the measured pillar along its central axis, the height _sides, the height of the pillar measured along its sides at the highest point on the edge. The _maximum height is the height from the bottom of the pillar to the highest position on its top. The _maximum height is the normalization coefficient.

The TIR of individual copper pillars may be determined by the following equation:
TIR = height _{center -height side}
The height _center and height _sides are as defined above.

Pillar parameters for determining TIR and TIR% can be measured using optical profileometry, such as using a white light LEICA DCM 3D or similar device.

The copper pillars of the present invention further include a tin or tin alloy solder bump 108 and a copper-tin metal interface 116. The copper-tin metal interface joins the bottom section 112 of the solder bump 108 to the top 106 of the copper pillar to form a complete copper pillar 100 structure.

The copper-tin metal interface 116 has a length 120 and a thickness or width from one end to the other end of the copper-tin metal interface, where most of the voids are concentrated and the thickness is preferably. It is in the range of 1 μm to 5 μm, more preferably 1 μm to 3 μm, even more preferably 1 μm to 2 μm, and most preferably 1 μm. The width can be measured using conventional methods well known in the art. The copper pillars of the present invention form voids after reflow of 20% or less, preferably 15% or less, more preferably 0.5% to 15%, most preferably 0% to 5% along the length and width. Has.

The percentage of void formation after reflow is determined by the following equation.
V% = total length / pillar diameter (μm) of the number of voids counted along the length (μm) of the interface and within the width of the interface, from copper to copper-tin metal in the copper pillar to 1-5 μm. ) × 100

The length of the copper-tin metal interface 120 is parallel to the pillar diameter on the central axis of the pillar and can be equal to the pillar diameter. When the top of the pillar is dish-shaped or dome-shaped, the length of the copper-tin metal interface for determining V% is the length of the ellipse. Conventional equations well known in the art and conventional methods for determining the length of an ellipse can be used.

The number of voids and their length (diameter) include, but are not limited to, ImageJ (available at https://imagej.nih.gov/ij/) available from NIH, Bethesda, Maryland, USA. It is counted by a conventional counting program such as software used for images. NIH's ImageJ is a public domain Java®-based image processing program developed at NIH. ImageJ is designed with an open architecture that provides extensibility with Java® plug-ins and recordable macros. You can use ImageJ's built-in editor and Java® compiler to develop custom acquisition, analysis, and processing plug-ins. Voids can range in length (diameter) from less than 0.1 μm to more than 0.5 μm. Copper-tin metal because many small length (diameter) adjacent voids in the interface coalesce to form a much larger void, which can cause copper pillar breakage at the copper-tin metal interface. The total length of voids in the copper-tin metal interface along the length and width of the interface (1-5 μm to the interface) is important.

FIG. 3 shows the copper-tin metal interface 116 of the present invention showing an interface along length 120 having V% = substantially 0% in the absence of detectable voids. In contrast, FIG. 4 shows a conventional copper-tin metal interface outside the scope of the present invention, which contains equivalent amounts of voids 117 of various diameters along the length of the interface 116.

Although the present invention has substantially described a method of electroplating copper pillars having a columnar morphology, the copper pillars can be oval, octagonal and rectangular, in addition to, for example, columnar or cylindrical. The method of the present invention is preferably for electroplating columnar copper pillars. Preferably, the columnar copper pillars have a flat or dish-shaped (concave) top. Preferably, the copper pillars of the present invention have an aspect ratio such as 3: 1 to 1: 1 or 2: 1 to 1: 1.

The substrate is electroplated by bringing the substrate into contact with the plating bath. The substrate functions as a cathode. The plating bath comprises an anode, which may be soluble or insoluble. An electric potential is applied to the electrodes. The overall average current density for electroplating a complete copper pillar is 0.25 ASD-40 ASD, preferably 1 ASD-30 ASD, more preferably 10 A.
It may be in the range of SD to 30 ASD, most preferably 10 ASD to 20 ASD.

The copper pillars 200 and 300 of the present invention can be formed by first depositing conductive seed layers (not shown) on substrates 202 and 302 such as semiconductor chips or dies to form an array of copper pillars. can. The substrate is then coated with photoresist materials 204 and 304. The photoresist layer can be applied to the surface of a semiconductor chip by a conventional process known in the art. The thickness of the photoresist layer can vary depending on the height of the pillars. The thickness range is preferably 40 μm to 250 μm, more preferably 40 μm to 50 μm. A patterned mask is applied to the surface of the photoresist layer (not shown) and imaged to selectively expose the photoresist layer to radiation such as UV irradiation (not shown). The photoresist layer can be a positive or negative photoresist. When the photoresist is of the positive type, the portion of the photoresist exposed to radiation is removed with a developer such as an alkaline developer. A pattern of multiple openings 206 and 306, such as vias, is formed on the surface that extends completely to the seed layer on the substrate. The diameter of the via can vary depending on the diameter of the pillar. The diameter of the via can be in the range of 2 μm to 300 μm, preferably 50 μm to 225 μm. The entire structure of the substrate and the developed photoresist with openings is then placed in a copper electroplating bath (not shown).

The first section of each copper pillar is electroplated in the initial high current density range. Preferably, the high current density range is greater than 10 ASD, preferably 15 ASD to 30 ASD, more preferably 15 ASD to 25 ASD, and most preferably 15 ASD to 20 ASD. A high current density is applied to the height of the first 30 μm to 35 μm of each copper pillar, followed by electroplating the second section or the rest of the height of the copper pillar with a current density lower than the initial current density. do. The lower current density is preferably 10 ASD or less, preferably 0.5 ASD to 10 ASD, more preferably 10 ASD to 1 ASD, and most preferably 8 ASD to 3 ASD. The second category of copper pillars plated with lower current densities is preferably in the height range of 1 μm to 10 μm, more preferably 1 μm to 5 μm, even more preferably 1 μm to 3 μm, most preferably 1 μm to 2 μm. .. 5A-5D and 6A-6B show methods of electroplating flat top copper pillars and dish-shaped copper pillars, respectively, while the method of the invention comprises dome-shaped (convex) copper pillars. It can also be used for plating. Preferably, the copper pillar of the present invention is a copper pillar having a flat top or a dish-shaped top. Most preferably, the copper pillars are dish-shaped (concave).

The copper electroplating bath used to plate the copper pillars of the present invention is one or more copper ion sources, one or more acids, one or more chloride sources, one or more levelers, one or more. Accelerators, one or more inhibitors and one or more additional components commonly found in copper electroplating baths, and water as a solvent. Such copper electroplating baths have an acidic pH range. Preferably, the pH of the copper electroplating bath is 2 or less, more preferably 1 or less, and most preferably less than 1. The same copper electroplating bath can be used to plate the entire copper pillar, or the copper electroplating bath used to plate the second section of the copper pillar or the rest of the copper pillar is copper. It may be a copper electroplating bath containing the minimum number of components for plating the pillars. Such a copper electroplating bath provides water, one or more copper ion sources, an electrolyte and 1 for maintaining an acidic pH, preferably 2 or less, more preferably 1 or less, most preferably less than 1. It consists of more than one kind of acid and optionally one or more chloride sources, one or more levelers, one or more accelerators and one or more inhibitors. Preferably, the second or alternative copper electroplating bath provides water, one or more copper ion sources, and an electrolyte, having an acidic pH, preferably 2 or less, more preferably 1 or less, most preferably less than 1. Consists of one or more acids, for maintaining. Examples of commercially available copper plating baths are INTERVIA ™ 8540, 9000 and 9600 copper electroplating baths available from Dow Electrical Materials in Marlborough, Massachusetts.

Next, tin or tin alloy solder bumps are deposited on the top of each copper pillar. Tin or tin alloy solder can be deposited on the top of copper pillars by conventional methods such as electroplating or chemical and physical deposition. Preferably, tin-tin alloy solder is deposited by electroplating solder bumps on the top of the copper pillars. The overall structure of the substrates 202 and 302 with the copper pillars 200 and 300 and the photoresists 204 and 304 is then a plating bath containing a solder such as tin solder or a tin alloy solder such as tin-silver or tin-lead alloy. Moved to. Solder bumps 205 and 305 are electroplated onto a substantially flat or dish-shaped surface of each copper pillar to fill a portion of the via. After plating the solder bumps, the photoresist is stripped from the substrate using a solvent stripper or by other conventional means known in the art to place an array of copper pillars with solder bumps on the substrate. leave. The rest of the seed layer not covered by the pillars is removed by an etching process well known in the art. The copper pillar with solder bumps and the substrate are then reflowed in a conventional reflow oven. An example of a reflow oven is Sikiama International, Inc., which includes five heating zones and two cooling zones. FALCON 8500 tool. The reflow cycle can be in the range 1-5. The reflow process results in the formation of copper-tin intermetallic interfaces 208 and 308, where V% is 20% or less, preferably 15% or less, more preferably 0.5% to 15%, most preferably. Is 0% to 5%.

Copper pillars with solder bumps come into contact with metal contacts on a printed circuit board (not shown), another wafer or die, or a substrate such as an interposer (not shown) that can be made of organic laminate, silicon, or glass. Is placed. The solder bumps are heated by a conventional process known in the art to reflow the solder and join the copper pillars to the metal contacts (not shown) on the substrate. Conventional reflow processes for reflowing solder bumps, such as the Sikiama International, Inc. FALCON 8500 tool reflow oven described above, can be used. The copper pillars are in physical and electrical contact with the metal contacts of the substrate. The underfill material can then be injected to fill the space between the die, pillars, and substrate. Conventional underfills well known in the art can be used.

An underbump metallized layer typically composed of a material such as titanium, titanium tungsten, or chromium is die to provide metal contacts and adhesions between the copper pillars and the semiconductor die during electroplating of the pillars. Precipitated on top. Alternatively, a metal seed layer such as a copper seed layer can be deposited on the semiconductor die to provide a metal contact between the copper pillar and the semiconductor die. After the photosensitive layer is removed from the die, all parts of the underbump metallized layer or seed layer are removed except for the lower part of the pillars. Conventional processes known in the art can be used.

The aqueous copper electroplating bath of the present invention comprises one or more copper ion sources, one or more acids, one or more accelerators (brighteners), one or more inhibitors, one or more levelers, and Optionally comprises one or more halide ion sources, preferably chlorides, and water. Additional optional ingredients such as buffers and antibacterial agents can be included. Such additional optional ingredients are conventional and well known to those of skill in the art. Many of such ingredients are readily available on the market. Alternatively, the aqueous copper electroplating bath is one or more copper ion sources, one or more acids, water, and optionally one or more accelerators, one or more inhibitors, one or more levelers, one. It comprises the above-mentioned halide ion source, one or more kinds of buffering agents, and one or more kinds of antibacterial agents. Preferably, the alternative aqueous copper electroplating bath comprises one or more copper ion sources, one or more acids and water. It contains virtually no additional ingredients. An alternative or second aqueous copper electroplating bath is used to plate the second section of copper pillars as described above.

Copper ion sources include copper halides such as copper sulfate and copper chloride, copper acetate, copper nitrate, copper tetrafluoroborate, copper alkylsulfonate, copper arylsulfonate, copper sulfamate, copper perchlorate, and glucon. Examples include, but are not limited to, copper acid acid. Exemplary copper alkane sulfonates include copper (C _1- C ₆ ) alkane sulfonate, and more preferably copper (C _1- C ₃ ) alkane sulfonate. Preferred copper alkanesulfonates are copper methanesulfonate, copper ethanesulfonate, and copper propanesulfonate. Copper aryl sulfonate includes, but is not limited to, copper benzene sulfonate and copper p-toluene sulfonate. A mixture of copper ion sources may be used. One or more salts of metal ions other than copper ions can be added to the electroplating bath of the present invention. Preferably, the copper salt is present in an amount sufficient to provide copper ions at a plating solution concentration of 30 g / L to 70 g / L. More preferably, the amount of copper ions is 40-60 g / L.

To provide an acidic copper plating bath, one or more acids are included in the aqueous copper electroplating bath. Preferably, the pH of the copper electroplating bath is 2 or less, more preferably 1 or less, and most preferably less than 1. Examples of the acid include alkane sulfonic acid such as sulfuric acid, acetic acid, fluoroboric acid, methane sulfonic acid, ethane sulfonic acid, propane sulfonic acid, and trifluoromethane sulfonic acid, and aryl sulfonic acid such as benzene sulfonic acid, p-toluene sulfonic acid, and sulfamic acid. Examples include, but are not limited to, acids, hydrochloric acids, hydrobromic acids, perchloric acids, nitrates, chromic acids, and phosphoric acids. Preferred acids include sulfuric acid, methanesulfonic acid, ethanesulfonic acid, propanesulfonic acid, hydrochloric acid, and mixtures thereof. The acid can be present at a concentration of 1 g / L to 400 g / L. Such acids are generally commercially available from a variety of sources.

Optionally, one or more halide ion sources can be included in the aqueous copper electroplating bath of the present invention. One or more sources of chloride and bromide ions can be included in the copper electroplating bath. Chloride ion sources include, but are not limited to, copper chloride, sodium chloride, potassium chloride, and hydrochloric acid. Sources of bromide ions include, but are not limited to, sodium bromide, potassium bromide, and hydrogen bromide. Preferably, the halide ion is a chloride. The halide ion concentration can be in the range of 0 to 200 ppm, preferably 50 ppm to 150 ppm, more preferably 60 ppm to 100 ppm, based on the plating bath. Halide ion sources are generally commercially available.

As accelerators, N, N-dimethyl-dithiocarbamic acid- (3-sulfopropyl) ester, 3-mercapto-propylsulfonic acid- (3-sulfopropyl) ester, 3-mercapto-propylsulfonic acid sodium salt, 3- Carbonate with potassium mercapto-1-propanesulfonic acid, dithio-O-ethyl ester-S-ester, bis-sulfopropyl disulfide, bis- (sodium sulfopropyl) -disulfide, 3- (benzothiazolyl-S-thio) propyl Sulfonic acid sodium salt, pyridinium propyl sulfobetaine, 1-sodium-3-mercaptopropan-1-sulfonate, N, N-dimethyl-dithiocarbamic acid- (3-sulfoethyl) ester, 3-mercapto-ethylpropyl sulfonic acid- (3) -Sulfoethyl) ester, 3-mercapto-ethyl sulfonic acid sodium salt, 3-mercapto-1-ethanesulfonic acid potassium salt, carbonate-dithio-O-ethyl ester-S-ester, bis-sulfoethyl disulfide, 3-( Examples include, but are not limited to, benzothiazolyl-S-thio) ethyl sulfonic acid sodium salt, pyridinium ethyl sulfobetaine, and 1-sodium-3-mercaptoethane-1-sulfonate. The accelerator can be included in the aqueous copper electroplating bath in an amount of 0.1 ppm to 1,000 ppm, preferably 0.5 ppm to 500 ppm, more preferably 0.5 ppm to 100 ppm.

Inhibitors include, but are not limited to, polypropylene glycol copolymers and polyethylene glycol copolymers, including, but not limited to, ethylene oxide-propylene oxide (“EO / PO”) copolymers and butyl alcohol-ethylene oxide-propylene oxide copolymers. The weight average molecular weight of the inhibitor can range from 800 to 15,000, preferably 1,000 to 15,000. When such inhibitors are used, they are preferably present at concentrations of 0.5 g / L to 15 g / L, more preferably 0.5 g / L to 5 g / L, based on the weight of the composition.

The aqueous copper electroplating bath of the present invention may include various levelers, preferably one or more amine compounds and one or more polyepoxides as disclosed in US2007 / 0007143. It is a reaction product, the entire disclosure of which is incorporated herein by reference. The amine compound is preferably a heterocyclic nitrogen compound, more preferably the heterocyclic nitrogen compound is imidazole, and the polyepoxide is preferably 1,4-butanediol diglycidyl ether, ethylene glycol diglycidyl ether, di. (Ethethylene glycol) Diglycidyl ether, glycerol diglycidyl ether, neopentyl glycol diglycidyl ether, 1,3-butanediol diglycidyl ether, propylene glycol diglycidyl ether, di (propylene glycol) diglycidyl ether, poly (ethylene glycol) It is selected from diglycidyl ether compounds and poly (propylene glycol) diglycidyl ether compounds. Most preferably, the polyepoxide is selected from glycerol diglycidyl ethers and neopentyl glycol diglycidyl ethers.

Preferably, the leveler can be prepared by dissolving the desired amount of one or more amines in water and heating the solution to about 40-90 ° C. with stirring. One or more polyepoxide compounds are then added to the solution with continued stirring. After adding the polyepoxide compound, the reaction mixture is heated to about 75-95 ° C. for about 4-8 hours. After stirring for 12-18 hours, the reaction mixture is diluted with water to adjust the pH to about 7.

The ratio of one or more amine compounds used to prepare the leveler to one or more polyepoxide compounds can be 0.1:10 to 10:01. Preferably, this ratio is 0.5: 5 to 5: 0.5, more preferably 0.5: 1 to 1: 0.5.

Preferably, the leveler has a number average molecular weight (Mn) of 1,000 to 10,000. Preferably, the leveler has a weight average molecular weight of 1,000 to 50,000, more preferably 1,000 to 20,000, even more preferably 1,500 to 5,000, or 5,000 to 15,000. It has a Mw) value.

The amount of leveler used in the copper electroplating bath can range from 0.25 ppm to 1,000 ppm, preferably 0.5 ppm to 500 ppm, more preferably 5 ppm to 500 ppm, based on the total weight of the plating bath.

The aqueous copper electroplating bath can also contain other leveling agents. Such leveling agents include US Pat. No. 6,610,192 by Step et al., US Pat. No. 7,128,822 by Wang et al., And US Pat. No. 6,800,188 by Hagiwara et al., And the United States. Examples thereof include those disclosed in Patent Publication Nos. 2017/004203037, 2017/0037526, 2017/0037527 and 2017/0037528, but the contents of these disclosures are not limited thereto. The whole is incorporated herein by reference.

The water-based copper electroplating bath can be used at a temperature of 10 to 65 ° C. or higher. Preferably, the temperature of the plating composition is 15-50 ° C, more preferably 20-40 ° C.

Preferably, the copper electroplating bath is agitated during use. Any suitable stirring method can be used, and such methods are well known in the art. Suitable agitation methods include, but are not limited to, air sparging, workpiece agitation, and impingement.

Tin or tin alloy solder can be deposited on top of copper pillars by conventional methods well known in the art such as electroplating and chemical and physical deposition methods, but tin or tin alloy solder is preferred. Is electroplated on the top of the copper pillar, more preferably a tin-silver alloy is electroplated on the top of the copper pillar. Conventional water-based tin-silver electroplating baths and tin-silver solder electroplating methods well known to those skilled in the art can be used. Examples of commercially available tin alloy electroplating baths are the SOLDERON ™ TS4000 and TS6000 tin-silver alloy electroplating baths available from Dow Electrical Materials in Marlborough, Massachusetts.

A preferred aqueous tin-silver electroplating bath comprises one or more tin ion sources selected from tin halide, tin sulfate, tin alkanesulfonate, tin alkanolsulfonate and the like, and an acid. One or more silver ion sources are selected from silver halide, silver gluconate, silver citrate, silver lactate, silver nitrate, silver sulfate, silver alkane sulfonate and silver alkanol sulfonate. One or more tin salts are contained in the tin-silver alloy bath in an amount of 30 g / L to 100 g / L, preferably 50 g / L to 100 g / L. One or more silver salts are contained in the tin-silver electroplating bath in an amount of 0.01 g / L to 20 g / L, preferably 0.01 g / L to 15 g / L.

One or more acids are contained in the tin-silver alloy bath. Acids include alkane sulfonic acids such as aryl sulfonic acid, methane sulfonic acid, ethane sulfonic acid and propane sulfonic acid, aryl sulfonic acid such as phenyl sulfonic acid and trill sulfonic acid, and sulfuric acid, sulfamic acid, hydrochloric acid and hydrobromic acid. And inorganic acids such as fluorosulphonic acid, but not limited to these. Preferably, the acid is selected from alkane sulfonic acid and aryl sulfonic acid. One or more acids are contained in an amount of 0.01 g / L to 500 g / L, preferably 10 g / L to 400 g / L, more preferably 50 g / L to 200 g / L.

One or more dihydroxybis-sulfide compounds can be included in the tin-silver alloy bath. Such compounds are 2,4-dithia-1,5-pentanediol, 2,5-dithia-1,6-hexanediol, 2,6-dithia-1,7-heptanediol, 2,7-dithia. -1,8-octanediol, 2,8-dithia-1,9-nonanediol, 2,9-dithia-1,10-decanediol, 2,11-dithia-1,12-dodecanediol, 5,8 -Dithia-1,12-dodecanediol, 2,15-dithia-1,16-hexadecanediol, 2,21-dithia-1,2-dodecanediol, 3,5-dithia-1,7-heptanediol , 3,6-dithia-1,8-octanediol, 3,8-dithia-1,10-decanediol, 3,10-dithia-1,8-dodecanediol, 3,13-dithia-1,15- Pentadecanediol, 3,18-dithia-1,20-eicosanediol, 4,6-dithia-1,9-nonanediol, 4,7-dithia-1,10-decanediol, 4,11-dithia-1 , 14-Tetradecanediol, 4,15-dithia-1,18-octadecanediol, 4,19-dithia-1,2-dodecanediol, 5,7-dithia-1,11-undecanediol, 5,9 -Dithia-1,13-tridecanediol, 5,13-dithia-1,17-heptadecanediol, 5,17-dithia-1,21-uneicosandiol and 1,8-dimethyl-3,6- It is selected from dithia-1,8-octanediol. Such compounds are included in an amount of 0.1 g / L to 15 g / L, preferably 0.5 g / L to 10 g / L.

One or more mercaptotetrazole can be included in the tin-silver alloy bath. Such mercaptotetrazole is 1- (2-diethylaminoethyl) -5-mercapto-1,2,3,4-tetrazole, 1- (3-ureidophenyl) -5-mercaptotetrazole, 1-((3-) It is selected from N-ethyloxalamide) phenyl) -5-mercaptotetrazole, 1- (4-acetamidophenyl) -5-mercapto-tetrazole and 1- (4-carboxyphenyl) -5-mercaptotetrazole. The mercaptotetrazole compound may be included in the bath in an amount of 1 g / L to 200 g / L, preferably 5 g / L to 150 g / L.

Optionally, one or more inhibitors can be included in the tin-silver alloy bath. Such inhibitors include, but are not limited to, alkanolamines, polyethyleneimines and alkoxylated aromatic alcohols. Suitable alkanolamines include substituted or unsubstituted methoxylated, ethoxylated, and propoxylated amines such as tetra (2-hydroxypropyl) ethylenediamine, 2-{[2- (dimethylamino) ethyl] -methylamino. } Ethanol, N, N'-bis (2-hydroxyethyl) -ethylenediamine, 2- (2-aminoethylamine) -ethanol, and combinations thereof, but are not limited thereto. Preferably, they are contained in an amount of 0.5 g / L to 15 g / L, more preferably 1 g / L to 10 g / L.

Polyethyleneimine includes, but is not limited to, substituted or unsubstituted linear or branched chain polyethyleneimine having a molecular weight of 800 to 750,000 or a mixture thereof. Suitable substituents include, for example, carboxyalkyl, such as carboxymethyl, carboxyethyl.

Alkoxylated aromatic alcohols include, but are not limited to, ethoxylated bisphenols, ethoxylated betanaphthols, and ethoxylated nonylphenols.

Optionally, one or more reducing agents can be added to the bath to help keep the tin in a soluble divalent state. Reducing agents include, but are not limited to, hydroquinone, hydroquinone sulfonic acid, potassium salts, and hydroxylated aromatic compounds such as resorcinol and catechol. Such reducing agents are included in an amount of 0.01-20 g / L, preferably 0.1-5 g / L.

Optionally, one or more grain refiners can be included in the tin-silver alloy bath. Examples of such grain refiners include alkoxylates such as polyethoxylated amine JEFFAMINE T-403 or TRITON RW, or sulfated alkylethoxylates such as TRITON QS-15, and gelatin or gelatin derivatives. Not limited to these. Alkoxylated amine oxides may also be included. Various alkoxylated amine oxide surfactants are known, but low effervescent amine oxides are preferably used. Such preferred alkoxylated amine oxide surfactants have viscosities of less than 5,000 cps as measured using a Brookfield LVT viscous meter equipped with a # 2 spindle. Preferably, this viscosity is determined by the ambient temperature. Preferably, such grain refiners are included in an amount of 0.5 g / L to 20 g / L.

The tin-silver alloy bath is acidic. Preferably, the tin-silver alloy bath has a pH of less than 1-2, more preferably less than 1.

The tin-silver alloy is electroplated from 0.05 ASD to 25 ASD, preferably 0.05 ASD to 10 ASD.

The tin-silver alloy can be electroplated at room temperature to 55 ° C., preferably room temperature to 40 ° C., more preferably room temperature to 30 ° C.

The following examples are intended to further illustrate the invention and are not intended to limit the scope of the invention.

Examples 1-5 (Invention)
Copper Pillars with Reduced Void Formation Prepare an aqueous acidic copper electroplating bath with the components disclosed in Table 1 below.

Prepare an aqueous acidic tin-silver alloy electroplating bath for solder bumps with the components disclosed in Table 2 below.

Example 1:
A 50 μm thick patterned photoresist and a 300 mm silicon wafer segment with multiple openings (available from IMAT Inc., Vancouver, WA) is immersed in a copper electroplating bath. The anode is a soluble copper electrode. The wafer and anode are connected to a rectifier and multiple copper pillars are electroplated on an exposed metal seed layer at the bottom of the circular opening that allows the formation of pillars with columnar morphology. The opening diameter is 50 μm to allow the formation of pillars with an average diameter of 45 μm. The temperature of the copper electroplating bath is 25 ° C. during plating. The initial average current density during plating of the first 33 μm (first vertical section) of the copper pillar is 20 ASD, followed by an average current density of 0. Reduce to 5 ASD. The overall average current density for electroplating copper pillars to their final height of 35 μm is 14.9 ASD.

Pillar height and TIR are measured using an optical white light LEICA DCM 3D microscope. TIR% is determined by the following formula.
TIR% = [Height _center -Height _side ] / _Maximum height x 100
TIR = height _{center -height side}
The TIR% of the pillars is determined to be about -7.9%. The pillars have a dish-shaped or concave top shape.

Wafer segments with photoresists and copper pillars are then immersed in the tin-silver electroplating baths disclosed in Table 2. The anode is a soluble tin electrode. The wafer and anode are connected to a rectifier and a 25 μm high tin-silver solder bump is plated on the top of each copper pillar. The tin-silver bath is 25 ° C. in plating and the average current density is 10 ASD.

After plating the tin-silver solder on the copper pillars, the photoresist is stripped with a BPR photo stripper solution available from the Dow Chemical Company, and an array of copper pillars with tin-silver solder bumps on the top of each pillar is placed on the wafer. Leave in.

Next, a wafer segment having copper pillars and tin-silver solder bumps was obtained from Sikiama International, Inc. Place in the FALCON 8500 tool reflow oven. Wafer segments in 5 reflow zones (140 ° C, 190 ° C, 190 ° C, 230 ° C and 260 ° C) for 30 seconds per zone (20 seconds transport and 10 seconds stationary), cooldown zone, and ramp speed 40 ° C / sec. Let it pass by.

The wafer segment is removed from the reflow oven and the V% is determined along the interface where the copper-tin metal section and the copper of the pillars meet. The average length of the copper-tin metal section for multiple copper pillars is 46.5 μm, and the width or thickness of the copper-tin metal section where voids are counted within the copper-to-copper-tin metal section is 1 μm. .. Counting and summing the length of each void along the length and width of the interface is done by ImageJ software available from NIH (see https://imagej.nih.gov/ij/).

The process steps for counting the number and length of each void using the ImageJ program are as follows:
1. 1. Receive a cross-sectional SEM image (on the central axis) of each copper pillar with solder bumps,
2. Open an image with the ImageJ program,
3. 3. Draw a line on the scale bar of the image (yellow line),
4. Go to "Analysis"-"Set Scale", enter "Known Distance" based on the scale and press "OK",
5. Right-click on the Fixed Length Line Tool and set the Desired Line Length to 45 μm (approximately equal to the length of the copper-tin metal section along the top of the copper pillar).
6. Select the "centered" option and press "OK",
7. Place the cursor in the center of the pillar directly below the Cu-Cu _{3 Sn IMC line,}
8. Select the "Rectangle" tool and draw a rectangle that captures the _{Cu-Cu 3 Sn IMC along the 45 μm line.}
9. Go to "Image" and "Crop", and 10. At this point, the void is magnified, a line is drawn along the length or diameter of the void, and the length or diameter of the void is measured.

The formula used to determine V% for one or more pillars is:
Total total length of voids counted along the length (μm) of the copper-tin metal interface and within 1 μm of width or thickness of the copper-tin metal interface / pillar diameter (μm) x 100

The V% of the plurality of pillars is determined to be about 12%. At the copper-tin metal interface, there are no observable signs of cracking or breakage in any of the copper pillars.

FIG. 7A is a 5,000 times SEM cross section at the center of one of the copper pillars with tin-silver solder bumps after reflow. The dish-shaped portion of the bottom of the SEM is the top of the copper pillar. The brighter top of the SEM is the copper-tin metal section. Dark spots at the copper-copper-tin metal interface or bond line of pillars are voids of relatively very short length.

Example 2:
The method disclosed in Example 1 above is repeated, except that the current density applied during plating of the last 2 μm (second vertical section) of the copper pillar is 4 ASD. The average plating rate of copper pillars is 16.3 ASD. The top of the copper pillar is then plated with tin-silver solder bumps from the tin-silver bath disclosed in Table 2. Copper pillars have a TIR% of about -7.5% when measured using an optical white light LEICA DCM 3D microscope. The top morphology is dish-shaped or concave. The photoresist is stripped and the array of copper pillars with tin-silver bumps is reflowed according to the method described above.

The void length of these copper pillars is determined using the ImageJ program described above. V% is only about 1%. At the copper-tin metal interface, there are no observable signs of cracking or breakage in any of the copper pillars.

FIG. 7B is a SEM 5,000 times the cross section at the center of one of the reflowed copper pillars. Similar to FIG. 7A, the dish-shaped portion at the bottom is the top of the copper pillar. The brighter top of the SEM is the copper-tin metal section. As determined by the ImageJ program, the copper pillars contain voids, but the length of the voids is very small and no voids are observed in FIG. 7B.

Example 3:
The method disclosed in Example 1 above is repeated, except that the current density applied during plating of the last 2 μm (second vertical section) of the copper pillar is 6 ASD. The average plating rate of copper pillars is 17.6 ASD. The top of the copper pillar is then plated with tin-silver solder bumps from the tin-silver bath disclosed in Table 2. Copper pillars have a TIR% of about -8%. The top of the pillar has a dish-shaped shape. The photoresist is stripped and the array of copper pillars with tin-silver bumps is reflowed according to the method described in Example 1 above.

The void length of these copper pillars is determined using the ImageJ software program. V% is 5%. At the copper-tin metal interface, there are no observable signs of cracking or breakage in any of the copper pillars.

FIG. 7C is a SEM 5,000 times the cross section at the center of one of the reflowed copper pillars. Similar to FIG. 7A, the dish-shaped portion at the bottom is the top of the copper pillar. The brighter top of the SEM is the copper-tin metal section. Some voids can be seen very close to the copper surface.

Example 4:
The method disclosed in Example 1 above is repeated, except that the current density applied during plating of the last 2 μm (second vertical section) of the copper pillar is 8 ASD. The average plating rate of copper pillars is 18.4 ASD. The top of the copper pillar is then plated with tin-silver solder bumps from the tin-silver bath disclosed in Table 2. Copper pillars have a TIR% of about -8.2%. The shape of the top is dish-shaped. The photoresist is stripped and the array of copper pillars with tin-silver bumps is reflowed according to the method described in Example 1 above.

The void length of these copper pillars is determined using the ImageJ software program. Similar to Example 3, V% is 5%. At the copper-tin metal interface, there are no observable signs of cracking or breakage in any of the copper pillars.

Example 5:
The method disclosed in Example 1 above is repeated, except that the current density applied during plating of the last 2 μm (second vertical section) of the copper pillar is 10 ASD. The average plating rate of copper pillars is 18.9 ASD. The top of the copper pillar is then plated with tin-silver solder bumps from the tin-silver bath disclosed in Table 2. Copper pillars have a TIR% or about -8.3%. The top of the copper pillar has a dish-shaped shape. The photoresist is stripped and the array of copper pillars with tin-silver bumps is reflowed according to the method described in Example 1 above.

The void lengths of these copper pillars were determined using the ImageJ program. V% was 14%. At the copper-tin metal interface, there are no observable signs of cracking or breakage in any of the copper pillars.

The table below summarizes the plating parameters and results of Examples 1-5.

Example 6 (Comparative Example)
Copper Pillars with High Void Formation at Copper-Tin Interface 50 μm thick patterned photoresist and 300 mm silicon wafer segments with multiple openings (available from IMAT, Inc., Vancouver, WA), Examples 1-5. Immerse in the copper electroplating bath shown in Table 1. The anode is a soluble copper electrode. The wafer and anode are connected to a rectifier and the copper pillars are electroplated on the exposed metal seed layer at the bottom of the circular opening which allows the formation of pillars with columnar morphology. The opening diameter is 50 μm to allow the formation of pillars with an average diameter of 45 μm. The temperature of the copper electroplating bath is 25 ° C. during plating. The average current density during plating of the 35 μm high copper pillars is 20 ASD without any reduction in the average current density as in Examples 1-5 above.

Wafer segments with photoresists and copper pillars are then immersed in the tin-silver electroplating baths disclosed in Table 2. The anode is a soluble tin electrode. The wafer and anode are connected to a rectifier and a 25 μm high tin-silver solder bump is plated on the top of each copper pillar. The tin-silver bath is at 25 ° C. during plating and has an average current density of 10 ASD.

After plating the tin-silver solder on the copper pillars, the photoresist is stripped with a BPR photo stripper solution available from the Dow Chemical Company, and an array of copper pillars with tin-silver solder bumps on the top of each pillar is placed on the wafer. Leave in. The TIR% of the pillars is -9.6%. The top of each pillar has a dish-shaped shape.

Next, the wafer segment with copper pillars and tin-silver solder bumps is placed in a FALCON 8500 tool reflow oven from Sikiama International, Inc. Wafer segments in 5 reflow zones (140 ° C, 190 ° C, 190 ° C, 230 ° C and 260 ° C) for 30 seconds per zone (20 seconds transport and 10 seconds stationary), cooldown zone, and ramp speed 40 ° C / sec. Let it pass by.

The wafer segment is removed from the reflow oven and the V% is determined along the interface where the copper-tin metal section and the copper of the pillars meet at a thickness of 1 μm. Counting the length of each void along the interface and summing the lengths is done by ImageJ software. V% is 40%. This is a significant increase in voids compared to the much lower V% of the invention in Examples 1-5. FIG. 8 is an SEM 5,000 times the cross section at the center of one of the reflowed dish-shaped copper pillars. The SEM shows void formation along the bond line and voids of various lengths (diameters). Most copper pillars showed cracks along the copper-tin interface with tin-silver solder bumps.

Claims (2)

A copper pillar comprising a horizontal base and a vertical section that includes a bottom side and a top side opposite to the bottom side, wherein the bottom side of the vertical section is joined to the horizontal base and is tin or Tin alloy solder bumps are joined to the top side of the vertical section by a layer of copper-tin metallized compound, where V% is (along the length of the layer of copper-tin metallized compound and Total length of void count / pillar diameter) x 100 within the width from the interface between the copper of the copper pillar and the layer of the copper-tin metal compound to 5 μm within the layer of the copper-tin metal compound. There is a V% = 0.5% ~ 20% ,
The TIR% of the copper pillar is in the range of -9% to -8.2%.
The TIR% of the copper pillar is determined by the following formula.
TIR% = [Height _center -Height _side ] / _Maximum height x 100
In the formula, the _center of height is the height of the copper pillar measured along the central axis of the copper pillar, and the height _side is the highest on the side measured along the side of the copper pillar. The height of the copper pillar at a point, the _maximum height of which is the height from the bottom of the copper pillar to the highest position of the top of the copper pillar. The copper pillar according to claim 1, wherein the V% is 0.5% to 15%.

Images